Amir Zur

Hi there! I am an NLP researcher passionate about understanding how AI models learn and process language.

I completed my MS in Computer Science and BA in Linguistics at Stanford University, where I conducted research on causally-motivated explanation and evaluation of language models advised by Dr. Christopher Potts and Dr. Atticus Geiger.

My Honors thesis at Stanford focused on interpretable, de-biased, and accessible language models. At the Pr(Ai)²R Group, I researched the decision-making of language models during text generation. Right now, I am a data scientist at Microsoft working on interactive and explainable agents.

news

Sep 20, 2024	Our paper was accepted into EMNLP 2024 main conference! Excited to chat in Miami 😀
Aug 26, 2024	Started full-time as a Data Scientist at Microsoft!
Jun 12, 2024	Posted our paper Updating CLIP to Prefer Descriptions Over Captions on arxiv.

selected publications

EMNLP

Updating CLIP to Prefer Descriptions Over Captions

Amir Zur, Elisa Kreiss, Karel D’Oosterlinck, and 2 more authors

In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, Nov 2024

Bib PDF Code

@inproceedings{zur2024updating,
  title = {Updating CLIP to Prefer Descriptions Over Captions},
  author = {Zur, Amir and Kreiss, Elisa and D'Oosterlinck, Karel and Potts, Christopher and Geiger, Atticus},
  editor = {Al-Onaizan, Yaser and Bansal, Mohit and Chen, Yun-Nung},
  booktitle = {Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing},
  month = nov,
  year = {2024},
  address = {Miami, Florida, USA},
  publisher = {Association for Computational Linguistics},
  url = {https://aclanthology.org/2024.emnlp-main.1125},
  pages = {20178--20187},
}

JMLR

Causal Abstraction: A Theoretical Foundation for Mechanistic Interpretability

Atticus Geiger, Duligur Ibeling, Amir Zur, and 8 more authors

Nov 2024

Bib PDF

@article{geiger2024causalabstractiontheoreticalfoundation,
  title = {Causal Abstraction: A Theoretical Foundation for Mechanistic Interpretability},
  author = {Geiger, Atticus and Ibeling, Duligur and Zur, Amir and Chaudhary, Maheep and Chauhan, Sonakshi and Huang, Jing and Arora, Aryaman and Wu, Zhengxuan and Goodman, Noah and Potts, Christopher and Icard, Thomas},
  year = {2024},
  eprint = {2301.04709},
  archiveprefix = {arXiv},
  primaryclass = {cs.AI},
  url = {https://arxiv.org/abs/2301.04709},
}